RIBRush

Who it is for.

RIBRush is a browser puzzle for people learning how routers choose a next hop. Players route 35 packets by comparing destinations with IPv4 and IPv6 candidate routes, then get an explanation of the deciding rule. The seven levels build from simple longest-prefix decisions to Cisco administrative distance and same-protocol metrics, with a specific-route-versus-trusted-route trap. It is designed for early routing practice, not as a simulation of a vendor's full route-installation pipeline.

How it was made.

The project began with a constitution and a feature specification, then added research, a plan, data-model and UI contracts, a quickstart, and ordered implementation tasks. The route rules were grounded in Cisco IOS/IOS XE administrative-distance documentation and the relevant IPv4 and IPv6 longest-prefix references. The implementation separates address handling, route evaluation, deterministic scenario generation, run state, and presentation so the evaluator remains the authority for each answer. Human feedback shaped the project name, the Cisco AD baseline, the progression from IPv4 into IPv6, and the decision to mix LPM, AD, and metric warmups in Level 1. Repeated destination blocks across levels were addressed with distinct IPv4 route pools, and the specification and design documents were synchronized with those changes. The browser check page reports 12 passing checks, and a full 35-packet run was verified. The five-person formative playtest is still outstanding.

For the next builder.

• Keep the learning model explicit: LPM first, AD only for identical prefixes,
• then metric only for the same protocol and prefix.
• Preserve seeded generation and independently expected fixtures when adding
• route pools; varied data still needs reproducible correctness checks.
• Complete the five-person playtest and record evidence before calling the
• release criteria finished.